Using Proximity andHomophily to ConnectConference Attendees ina Mobile Social NetworkAlvin ChinMobile Social Experiences T...
Outline    • Motivation and research problem    • Contributions    • Find & Connect @ UbiComp 2011    • User behaviour ana...
Motivation    • Who should I meet at the conference?    • Who is this person that I met?    • Why should I add this person...
Homophily    • Social selection    • We connect with people who are similar to us as      friends       (McPherson et al, ...
Proximity    • Using location and human mobility for friendship (Cho       et al, 2011)    • Encounters to determine who t...
Drawback    • Fail to help users create and maintain their social      network at the same time to bring convenience and  ...
Research problem    • Determine how to use proximity and homophily to      connect attendees in a conferenceNokia Research...
Offline Encounters Influences Online FriendshipSource: Xu et al. Social Linking and Physical Proximity in a Mobile Locatio...
Offline Improves Friend RecommendationSource: Xu et al. Using Physical Context in a Mobile Social Networking Application f...
Recording offline interactions as ephemeralsocial networks                  Offline physical activities (Conf., Meeting, P...
Contributions     • Create Find & Connect, a platform combining the       conference program with indoor location and prox...
Find & Connect @ UbiComp 2011     • Allow conference attendees to connect with each other       during the conference base...
Find & Connect system                                       RFID positioning            Find & Connect   Mobile device RFI...
Find someone nearby during sessionNokia Research Center14                        Company Confidential
Find who this person is and what youhave in commonNokia Research Center15                        Company Confidential
Add this person as contactNokia Research Center16                        Company Confidential   16
See the conference program and whoattended the sessionsNokia Research Center17                        Company Confidential...
See notifications of who added you ascontactNokia Research Center                        Company Confidential   18
User behaviour analysis     • Demographics     • Feature usage     • Online connections     • Offline encountersNokia Rese...
Demographics     • Sept. 17 to 21, 2011 at Tsinghua University     • Workshops, tutorials, research      papers, posters, ...
Feature usage     • Finding people nearby (11.66%)     • Notices (10.30%)     • Login (6.27%)     • Program (4.97%)     • ...
Online connections: contacts graphNokia Research Center22                        Company Confidential
Online connections: contactsNokia Research Center23                        Company Confidential
Contacts degree distributionNokia Research Center24                        Company Confidential
Offline is the reason why people add friendsNokia Research Center25                        Company Confidential
Contact recommendation     • Weight vector wi :                 wi = {wci, wcf, wcs,we, | wci + wcf + wcs +we = 1, 0 < wf ...
Contact recommendations results     • 15252 total, 309 of them added by       63 users = 2% of all contact       recommend...
Offline connections: encounters graphNokia Research Center28                        Company Confidential
Offline connections: encounters     • 12,716,349 total encountersNokia Research Center29                           Company...
Encounters degree distributionNokia Research Center30                        Company Confidential
Implications     • Find & Connect can help people build connections in a       conference     • People add others as frien...
Conclusions     • Contact and encounter networks follow social influence theory of       3 degrees of separation       Cac...
Future work     • Improve user interface, users can post to online SNS       and can add friends to SNS     • Study relati...
Alvin Chin     Nokia Research Center, Beijing     alvin.chin@nokia.com     http://research.nokia.com/people/alvin_chin    ...
Upcoming SlideShare
Loading in …5
×

Using Proximity and Homophily to Connect Conference Attendees in a Mobile Social Network

1,206 views
1,120 views

Published on

Published in: Technology, Education
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,206
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
11
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Using Proximity and Homophily to Connect Conference Attendees in a Mobile Social Network

  1. 1. Using Proximity andHomophily to ConnectConference Attendees ina Mobile Social NetworkAlvin ChinMobile Social Experiences TeamNRC Growth Economies LabNokia Research CenterNokia Research Center
  2. 2. Outline • Motivation and research problem • Contributions • Find & Connect @ UbiComp 2011 • User behaviour analysis • Implications • Conclusions and future workNokia Research Center2 Company Confidential
  3. 3. Motivation • Who should I meet at the conference? • Who is this person that I met? • Why should I add this person to my social network?Nokia Research Center3 Company Confidential
  4. 4. Homophily • Social selection • We connect with people who are similar to us as friends (McPherson et al, 2001) • User similarity using people, places, things (Guy et al, 2010)Nokia Research Center4 Company Confidential
  5. 5. Proximity • Using location and human mobility for friendship (Cho et al, 2011) • Encounters to determine who to add as friend (Aka- Aki; Quercia and Capra, 2009) • Introduce people and infer one’s social network (Eagle and Pentland, 2005) • Enhancing social interactions at conferences (Barrat et al, 2010)Nokia Research Center5 Company Confidential
  6. 6. Drawback • Fail to help users create and maintain their social network at the same time to bring convenience and facilities to usersNokia Research Center6 Company Confidential
  7. 7. Research problem • Determine how to use proximity and homophily to connect attendees in a conferenceNokia Research Center7 Company Confidential
  8. 8. Offline Encounters Influences Online FriendshipSource: Xu et al. Social Linking and Physical Proximity in a Mobile Location-based Service, 1stInternational Workshop on Mobile Location-based Services, In Proc. of UbiComp 2011, 2011Nokia Research Center8 Company Confidential
  9. 9. Offline Improves Friend RecommendationSource: Xu et al. Using Physical Context in a Mobile Social Networking Application forImproving Friend Recommendations, 1st International Workshop on Sensing, Networking andComputing on Smartphones, In Proc. of CPSCom 2011, 2011 Nokia Research Center 9 Company Confidential
  10. 10. Recording offline interactions as ephemeralsocial networks Offline physical activities (Conf., Meeting, Party, Shopping, Hiking…) Activity 1 Activity 3 Activity 2 Activity n … Online social networks (Facebook, Twitter, Weibo, Renren…)Nokia Research Center time10 Company Confidential
  11. 11. Contributions • Create Find & Connect, a platform combining the conference program with indoor location and proximity • Deploy Find & Connect to UbiComp 2011 conference • Understand user behaviour in conference using social network analysis, data mining and survey techniquesNokia Research Center11 Company Confidential
  12. 12. Find & Connect @ UbiComp 2011 • Allow conference attendees to connect with each other during the conference based on • their location • Their common research interests • the sessions that they have attended • the attendees that they have encountered over the course of the conference • Common friendsNokia Research Center12 Company Confidential
  13. 13. Find & Connect system RFID positioning Find & Connect Mobile device RFID badge RFID readers with LANDMARC server algorithmNokia Research Center13 Company Confidential
  14. 14. Find someone nearby during sessionNokia Research Center14 Company Confidential
  15. 15. Find who this person is and what youhave in commonNokia Research Center15 Company Confidential
  16. 16. Add this person as contactNokia Research Center16 Company Confidential 16
  17. 17. See the conference program and whoattended the sessionsNokia Research Center17 Company Confidential 17
  18. 18. See notifications of who added you ascontactNokia Research Center Company Confidential 18
  19. 19. User behaviour analysis • Demographics • Feature usage • Online connections • Offline encountersNokia Research Center19 Company Confidential
  20. 20. Demographics • Sept. 17 to 21, 2011 at Tsinghua University • Workshops, tutorials, research papers, posters, videos, demos • 421 registered attendees, 241 used Find & Connect (57%) • Apple device (31.34%), Google Chrome (23.85%), Android (22.12%), Firefox (9.08%), Internet Explorer (8.29%)Nokia Research Center20 Company Confidential
  21. 21. Feature usage • Finding people nearby (11.66%) • Notices (10.30%) • Login (6.27%) • Program (4.97%) • Finding people farther away (3.29%)Nokia Research Center21 Company Confidential
  22. 22. Online connections: contacts graphNokia Research Center22 Company Confidential
  23. 23. Online connections: contactsNokia Research Center23 Company Confidential
  24. 24. Contacts degree distributionNokia Research Center24 Company Confidential
  25. 25. Offline is the reason why people add friendsNokia Research Center25 Company Confidential
  26. 26. Contact recommendation • Weight vector wi : wi = {wci, wcf, wcs,we, | wci + wcf + wcs +we = 1, 0 < wf < 1} •Relevance vector Ri : Ri = {Rci, Rcf, Rcs, Re} • Relevance Rf Jaccard similarity of that feature f between Ui and U as Rf = | Nf (Ui ∩ U) | / |Nf (Ui U U) | • Recommended score FRi FRi = wi · Ri = {wci, wcf, wcs, we}·{Rci, Rcf, Rcs, Re, }TNokia Research Center26 Company Confidential
  27. 27. Contact recommendations results • 15252 total, 309 of them added by 63 users = 2% of all contact recommendations converted into contact requests • Low conversion rate probably due to few people using the recommendations featureNokia Research Center27 Company Confidential
  28. 28. Offline connections: encounters graphNokia Research Center28 Company Confidential
  29. 29. Offline connections: encounters • 12,716,349 total encountersNokia Research Center29 Company Confidential
  30. 30. Encounters degree distributionNokia Research Center30 Company Confidential
  31. 31. Implications • Find & Connect can help people build connections in a conference • People add others as friends/contacts if have physically met them • Recommendations need to be more visible in order to be useful • Post survey results show features were useful and user interface as averageNokia Research Center31 Company Confidential
  32. 32. Conclusions • Contact and encounter networks follow social influence theory of 3 degrees of separation Cacioppo, J.T., Fowler, J.H., and Christakis, N.A. Alone in the crowd: the structure and spread of loneliness in a large social network. Journal of Personality and Social Psychology, 97, 6 (2009), 977. • Users add others as contacts because of homophily and proximity • encounters • common sessions • common friends • People generally find Find & Connect useful somewhat easy to useNokia Research Center32 Company Confidential
  33. 33. Future work • Improve user interface, users can post to online SNS and can add friends to SNS • Study relationship between online and offline • Create model to identify groups of encounters that indicate activity-based social networks (ephemeral social networks)Nokia Research Center33 Company Confidential
  34. 34. Alvin Chin Nokia Research Center, Beijing alvin.chin@nokia.com http://research.nokia.com/people/alvin_chin Facebook: Alvin Chin (alvin.chin@utoronto.ca) LinkedIn: alvin.chin@nokia.com Twitter: gadgetman4u Sina Weibo: http://weibo.com/2106762242 (gadgetman) Foursquare: Alvin Chin (alvin.chin@nokia.com) Google+: ubiquitousdude@gmail.comNokia Research Center34 Company Confidential

×